The present invention discloses methods, systems, and tools for knowledge processing by building maps of ontological subjects for compositions or the contents accompanying a request for service by a user. In one embodiment of the invention the method assigns and calculates an ontological subject power spectrum to each composition or ontological subject. The resulting power spectrum and the adjacency matrix of the map are used to evaluate the merits of the compositions in the context of reference universes. It is also used as a research guiding tool for knowledge discovery. The invention serves knowledge seekers, knowledge creators, inventors, discoverer, as well as general public by assisting and guiding users to assess their work, identify their unknowns, optimize their research trajectory, and provide higher quality contents. The method and system, thereby, is instrumental in increasing the speed and efficiency of knowledge acquisition, discovery, retrieval, as well as faster learning and problem solving.
|
7. A computer implemented method, comprising an implementation using a portion or whole capacity of one or more non-transitory computer readable media with a set of instructions thereon, executable by one or more processing devices, configured, while being or is executed, for providing an ontological subject map of a universe of a composition comprising:
partitioning the composition to a plurality of partitions;
extracting at least some of ontological subjects of said at least one composition;
constructing, using one or more data processing or computing devices, one or more data structure corresponding to at least one matrix representing co-occurrences of some pairs of ontological subjects from said some of ontological subjects within certain proximities in the composition;
evaluating, using one or more data processing or computing devices, one or more indications of importance of said ontological subjects in said universe, wherein one of the one or more indications of importance is based on counts of said ontological subjects in said composition;
forming an association set for one or more of said extracted ontological subjects based on their co-occurrence counts with other ontological subjects and the value of each associated ontological subject, and storing them in a database;
constructing, using one or more data processing or computing devices, one or more data structure corresponding to the ontological subject map from said matrix configured to uniquely place each ontological subject in the map wherein each ontological subject is connected to at least one of said uniquely, excluding itself, assigned ontological subjects; and
processing at least one of said one or more data structures by one or more processing or computing devices or storing the at least one of said one or more data structure onto one or more non-transitory computer-readable storages.
1. One or more data structures, embedded on one or more non-transitory computer-readable media, representing a map for a plurality of ontological subjects, used in a composition of ontological subject to describe a universe, said one or more data structure are built by processing the composition, by one or more processing or computing devices, comprising:
evaluating, using one or more data processing or computing devices, one or more indications of importance of said ontological subjects in said universe, wherein one of the one or more indications of importance is based on counts of said ontological subjects in said composition;
selecting a number of ontological subjects of the composition based on value of one or more indications of importance of said ontological subjects in said composition;
finding associations for one or more of said selected ontological subjects by evaluating co-occurrences of a plurality of pairs of ontological subjects within certain proximities in the composition,
evaluating association values between one of said selected ontological subjects with other ontological subjects of the composition based on said co-occurrences of the some of the plurality of the pairs of ontological subjects, in which said one of said selected ontological subject is a member of the pairs, and value of the one or more indications of importance of said other oncological subjects,
building one or more data structures, using the values of said one or more importance factors of the ontological subjects of the composition, the association values between the ontological subjects of the composition, and the association sets of said each of selected ontological subjects of the composition, to represent the map configured to show:
a multilayer graph wherein each node of the graph represents an ontological subject and wherein at each layer of the graph a node is indicated as growing if is connected to another growing node from above layer and connected to at least one node in below layer and a node is indicated as dormant if the node is only connected to at least one growing node in their above layer
wherein at least one of said one or more data structures representing the map are stored in non-transitory computer readable memories and are further processed by one or more data processing or computing devices.
12. A computer implemented method, comprising an implementation using a portion or whole capacity of one or more non-transitory computer readable media with a set of instructions thereon, executable by one or more processing devices, configured, while being or is executed, for processing at least one composition of ontological subjects comprising:
getting the at least one composition of ontological subjects,
partitioning, using one or more data processing or computing devices, the at least one composition into a plurality of partitions;
extracting, using one or more data processing or computing devices, some of ontological subjects of said at least one composition;
constructing, using one or more data processing, storage devices, or computing devices, one or more numerical data structures corresponding to at least one matrix representing participation of said extracted ontological subjects into said partitions by having a non-zero value in the corresponding entries of the participation matrix, wherein each row of the participation matrix is representative of one of said extracted ontological subjects and each column of the participation matrix is representative of one of said partitions or vice versa; and
constructing one or more data structures, using one or more data processing, storage devices, or computing devices, corresponding to at least one co-occurrence matrix representing co-occurrences of each two of said some of ontological subjects in the plurality of partitions of the at least one composition, from the one or more data structures corresponding to said at least one participations matrix,
evaluating an importance score for the extracted ontological subject of the at least one composition based on one or more predefined criteria of importance,
evaluating numerically an association value for a selected number of pairs of ontological subjects, using the data of said at least one participation matrix and said matrix representing co-occurrences of the extracted ontological subjects in the plurality of partitions of the at least one composition,
indexing some or all of the extracted ontological subjects in such a way that each said ontological subject is indexed under most important ontological subjects, excluding under itself, that it has the highest association value with,
processing or storing the indexing data into one or more data structure which arrays.
2. The one or more data structures of
a. selecting a first set of ontological subject, having at least one member, and selecting a first associated sets of ontological subjects, each of said first associated sets having at least one member, wherein each of the first associated sets is corresponded to a member of the first set, wherein each member of each set of said first associated sets having association value greater than a predefined threshold with at least one member of the first set;
b. indexing the members of the first set and the members of said first associated sets in a multilayer index with indices, by an indexing method comprising;
i. indexing the members of the first set wherein indices are configured to show places of the members in the first layer of said multilayer index;
ii. indexing the members of the first associated sets in next layer of the index, wherein indices are configured to show places of the members in that layer and to indicate the association of a member of the first set to the members of the corresponded first associated set;
iii. marking an ontological subject in its current layer of the index as dormant, if said ontological subject has been indexed in previous layer and if an ontological subject is not indexed in previous layer but is a member of more than one of the first associated sets, mark said ontological subject as dormant everywhere that appear in its current layer of the index except in one place in its current layer of the index;
c. selecting a second set of ontological subject, having at least one member, from the ontological subjects of the first associated sets which are not marked as dormant, i.e. non-dormant, in the index, and select a second associated sets, each of said second associated sets having at least one member, wherein each of the second associated sets is corresponded to a member of the second set, wherein each member of each set of said second associated sets having association value greater than a predetermined threshold with at least one member of the second set;
d. indexing the members of the second associated sets in next layer of the index wherein indices are configured to show places of the members in their current layer and to indicate the association of a member of the second set to the members of corresponded second associated set, and marking an ontological subject in its current layer of the index as dormant, if said ontological subject has been indexed in one of the previous layers and if the ontological subject is not indexed in one of the previous layers but is a member of more than one of the second associated sets, mark the said ontological subject as dormant everywhere that appear in its current layer of the index except in one place in its current layer of the index; and
e. repeating the step c to form a next set of ontological subjects from the not marked as dormant, i.e. non-dormant, members of the last associated sets of ontological subject and to form a next associated sets and indexing the ontological subject of the said next associated sets by repeating the step d, wherein the term the second is replaced with the third, the forth and so on, whereby to uniquely index the desired number of ontological subjects in the context of said at least one composition, and wherein each ontological subject positioned once in the index as non-dormant; and
storing the data corresponding to the index of the ontological subject onto one or more non-transitory computer-readable storage devices.
3. The one or more data structures of
4. The one or more data structures of
5. The one or more data structures and the map of
6. The one or more data structures of
a. related ontological subjects;
b. most important subjects related to another ontological subject; and
c. new knowledge discovery or trajectory between two or more ontological subjects,
whereby increasing efficiency in searching, finding, and acquiring new knowledge.
8. The computer implemented method of
9. The computer implemented method of
a. providing one or more data structures, corresponding to a reference ontological subject index which corresponds to the ontological subject map of the reference universe, built from a selected list of ontological subjects and a selected collection of compositions;
b. building one or more data arrays corresponding to an ontological subject index of the universe of the received composition, wherein the index has at least one layer and has at least one ontological subject at its first layer;
c. scoring the merit of the received composition by a predetermined formula by processing the data of the ontological subject index of the received composition in view of the data of the reference ontological subject index.
10. The computer implemented method of
11. The computer implemented method of
constructing, using one or more data processing or computing devices, one or more data structures corresponding to at least one participation matrix representing participation of said extracted ontological subjects into said partitions by having a non-zero value in the corresponding entries of the participation matrix and a value of zero otherwise, wherein each row of the participation matrix is representative of one of said extracted ontological subjects and each column of the participation matrix is representative of said partitions or vice versa; and
building, using one or more data processing or computing devices, one or more data structures corresponding to at least one co-occurrence matrix from said at least one participation matrix.
13. The computer implemented method of
14. The computer implemented method of
15. The computer implemented method of
16. The computer implemented method of
17. The computer implemented method of
18. The computer implemented method of
19. The computer implemented method of
|
The present application is a divisional of U.S. patent application Ser. No. 12/547,879 filed on Aug. 26, 2009, which claims priority from U.S. provisional patent application Ser. No. 61/093,952 filed on Sep. 3, 2008, entitled “SYSTEM AND METHOD OF ONTOLOGICAL SUBJECT MAPPING FOR KNOWLEDGE PROCESSING APPLICATIONS” and also claims priority to the co-pending U.S. patent application Ser. No. 12/179,363 filed on Jul. 24, 2008 which claims priority from the Canadian patent application No. CA 2,595,541, filed on Jul. 26, 2007 entitled “ASSISTED KNOWLEDGE DISCOVERY AND PUBLICATION SYSTEM AND METHOD” which are incorporated entirely as references in this application.
This invention generally relates to information processing, knowledge processing and discovery, knowledge retrieval, artificial intelligence, ontology organization and applications.
Current methods of research and knowledge discovery utilizing informational retrieval systems and search engines are not very efficient. They are very time consuming and often requires that a user poses lots of expertise and previous knowledge and background to effectively use the information given by the informational retrieval systems about the user's subject matter of interest.
Currently, knowledge acquisition, retrieval, and discovery very much depend on the skill, dept and breadth of a researcher's knowledge. The problem that often slows down the knowledge acquisition and discovery is lack of awareness of unknowns. For example, when we want to do research in a subject or a topic, we usually use search engines to collect all the related data and often we overwhelmed with the number and extent of the documents found related to that subject. One should diligently read and analyze so many documents to find out that in order to master the subject/topic she/he should have known other subjects/topics first. Or find out after a long a period of time that in the process of understanding, analyzing and coming up with a solution, or a useful idea, related to a challenging problem, she/he has missed or overlooked many other important subjects. Therefore, it is important, desirable, and valuable to know and sort the most important things to know related to a subject.
In some other applications such as question answering and knowledge retrieval from a knowledge database, one need to know what are there to know at first and then looking for how they are related in order to build and enrich the knowledge database that is able to serve a client satisfactorily. Currently there is no systematic way of building such a database or general ontology consisting of statements of verified facts. The few attempts to build such useful databases are based on human labor or complicated reasoning and first order logic framework, rather than a systematic and automatic way of finding the distilled knowledge from the vast repositories of human generated data available over the internet.
More importantly in applications such as new knowledge discovery, education, essay examination, self education, scientific paper evaluation, composing new knowledge, business proposal evaluations, and many others it would be very helpful and desirable if we can measure the importance and intrinsic value of a composition in the universal context or in comparison with a large repository of knowledge. So that one can make sure that a composition is sound or the knowledge therein is genuinely valuable and original. Especially in publishing user generated content it is important to check the substance and merit of submitted contents before being published or sending through the costly process of expert reviewing.
All the above and many more arguments indicate a need in the art of knowledge discovery, knowledge retrieval, and knowledge publishing and the like, for a reference map of knowledge-bearer-components that is easy to navigate based on their intrinsic importance in the context of our universe. Moreover, there is a need for such knowledge maps and their corresponding databases for ranking and measuring the merit of newly composed documents or electronic contents and/or ranking existing compositions for more relevant retrieval, knowledge navigation, exploration, and generally assisting users in their research to gain speed and efficiency.
In this disclosure the knowledge-bearer-components are called Ontological Subjects (OS) and the present invention is about systems and methods of building ontological subject maps (OSMs). The system and method is instrumental for applications in ranking, merit evaluation of compositions, knowledge retrieval, knowledge exploration, research trajectory guiding, knowledge visualization, contribution measurement of new composition, and novelty detection as well as many other applications such as summarization, automatic report generation, question answering, and the like. The Ontological Subject Maps (OSM) are build using a plurality of ontological subjects by building the co-occurrence matrix and estimating association value of each two ontological subjects that have participated in one or more compositions or the parts thereof.
For each or any collection of compositions a universe, containing ontological subjects, is defined that the composition is trying to describe. According to one embodiment of the invention, the intrinsic importance of each ontological subject in the context of its universe is then calculated by defining a power value function for each ontological subject. The OSM can be build for single composition or for the entire repository of human knowledge such as the whole content of internet. The OSM and power spectrum of corresponding ontological subject build from the universe of the composition can be compared to a reference OSM build from a larger universe in order to assign a rank or merit for a composition for variety of applications mentioned above. Different exemplary metrics for comparison and merit valuation is proposed and introduced that are indicatives of intrinsic value of a composition such as authoritativeness, novelty, and substance. When the reference OSM is derived from a much larger repository or collection of repositories or the whole internet content, the calculated power of the ontological subjects are then universal. The power of OSs is, therefore, an indication of their intrinsic rule or importance in the real universe based on the comprehension of human beings since the start of civilization. Consequently ranking or assigning a weigh for each composition, based on its OS spectrum, is also universal.
In another embodiment, the reference OSM is proposed to be used for knowledge navigation and research trajectory identification. Since the map, when derived from large enough repositories, is basically map of human knowledge, a system and method is provided to guide a user to achieve her/his research goal much more efficiently and faster than using current search engine and knowledge navigation methods.
In yet another embodiment and application of the invention, the OSM is used to select a desirable number of OSs of interest and by way of searching and statistics to find the verified statements of the facts about that OS from a corpus or a collection of compositions related to that OS. Moreover for each two or more associated OS, it looks for explicitly expressed relations between those OSs and statistically verifies the specifics of their relationship and index the verified relationship in a knowledge database in the form of true statements containing two or more OSs. Thereby building a universal ontology becomes more automatic and efficient. The universal ontology has many important applications such as question answering and automatic useful knowledge discovery by reasoning and first order logic.
1. Ontological Subjects: means generally any string of characters, but more specifically letters, numbers, words, sound signal tracks, video signal tracks, or any combinations of them, and more specifically all such string combinations that indicates or refer to an entity, concept, quantities, and the incidences of such entities, concepts, and quantities. In this disclosure Ontological Subject/s and the abbreviation OS or OSs are used interchangeably.
2. Composition: means any combination of ontological subjects, particularly text documents written in natural language documents, data files, voice files, video files, and any mixture thereof.
3. Universe: in this disclosure “universe” is frequently used and have few intended interpretation: when “universe x” (x is a number or letter or word or combination thereof) is used it mean the universe of one or more compositions, that is called x, and contains none, one or more ontological subjects. By “real universe” or “our universe” we mean our real life universe including everything in it (physical and its notions and/or so called abstract and its notions) which is the largest universe intended and exist. Furthermore, “universal” refers to the real universe.
Furthermore the terms “storage device, “storage”, “memory”, and “computer-readable storage medium/media” refers to all types of no-transitory computer readable media such as magnetic cassettes, flash memories cards, digital video discs, random access memories (RAMSs), Bernoulli cartridges, optical memories, read only memories (ROMs), Solid state discs, and the like, with the sole exception being a transitory propagating signal.
All the electronic contents are compositions of a number of ontological subjects. The quality of composition's content in terms of its substance, validity, usefulness or beauty lies in the way that the content has been composed. While the number of possible compositions is endless, real life ontological subjects, however, are limited. All the compositions are talking about some of the ontological subjects of our universe. However, the real universe itself is one subject that has been functioning consistently for a long time. Therefore the underlying knowledge describing the working mechanism of our universe should be one well composed, e.g. written, composition that we as human are trying to uncover. Consequently the description of such system including all that there is in it and all the details should be straightforward once the relations between the parts become verifiably well known.
Our universe consists of parts, big and small, and whereas some parts of the universe is more important than the others. In describing our universe we usually can only focus on very small parts of it. However, focusing only on a small part of our universe can be misleading if the corresponding expressive statements don't get verified in the context of larger parts of the real universe. Therefore it is always more useful and more valuable to assess a composition, e.g. document, in larger context than a specific context or specific domain of discourse.
Universes of Compositions
In this invention we view each composition as a document that is trying to describe a universe of its own. The universe of a composition could be a true part of our real universe and is considered true if matches to a part of a larger part of the real universe, or could be partly true if it is not a perfect match and sometimes could be totally wrong if does not match at all. Currently the description of a universe by a composition is done by showing and establishing relations between ontological subjects of that universe by natural languages. Accordingly, we define, for each composition, a universe that the composition is trying to describe by combining and relating the ontological subjects of the corresponding universe.
It can be argued that, currently, the largest repository of human knowledge is the internet. A collection of billions of documents each has tried to prove or establish a relation between some parts of the real universe.
While it is difficult to become a master in any domain of discourse it becomes prohibitively difficult for human being to become a master in several domains of discourse. On the other hand more and more multidisciplinary expertise is required to discover important relations between the ontological subjects of different universes. Consequently it is important to know what are there to know in any universe and what is important to know firstly, secondly and so on. Hence whoever is trying to uncover some details of the real universe should be able to measure the importance and the value of his/her contribution in a much larger universal context. For instance, we have to have a way of distinguishing between a composition describing a valid and true breakthrough in cancer treatment from similar compositions claiming the same but in fact are partially valid and true. Furthermore, it is, for instance, important to know which discovery or invention is more important by how much and why. For example, discovery of new source of energy is more important than finding a solution for decreasing the production cost of a trivial specific product.
In a U.S. patent application entitled “Assisted Knowledge Discovery And Publication System And Method”, filed on Jul. 24, 2008, with the application Ser. No. 12/179,363 by the same inventor, an ontological subject mapping method was disclosed and an Ontological Subject Map (OSM) was used as a reference to assist in assessment of a submitted electronic content for considering for publication by an electronic publishing shop. In that application the preferred reference OSM is the universal OSM that is aiming to map all the possible and existing ontological subjects (OSs). However, such a universal map can take a long time to construct.
In one embodiment of the present invention we use a universe of reference which could be smaller than the universe of whole internet yet yield satisfactory benefits of a universal OSM. Accordingly we first introduce a method and system of building OSM for any exemplary universe of ontological subjects and then show the methods and systems of using such a map for different applications.
Participation, Co-Occurrence, and Association Value Matrixes for Building OSMs
Now we describe the steps of building an OSM for a composition describing universe 1, i.e. u1, in what follows here.
Break the composition to desired M number of partitions. For example, for a text document we can break the documents into chapters, pages, paragraphs, lines, and/or sentences etc. Identify the ontological subject of the composition by appropriate method such as parsing a text documents into its constituent words and phrases, and select a desired N number of the OS existing in the composition, according to certain predetermined criteria.
Then construct a N×M matrix in which the ith raw (Ri) is a vector, with dimension M, indicating the presence of OSi in each of the partitions of the composition (PC) by having a nonzero value, and not present by the value of zero. In most practical cases the matrix is a binary matrix, however in general said N×M matrix could assume any value as its entries.
We call this binary matrix the Participation Matrix (PM)
where PCi is the ith partitioned part of the composition, OSi is the ith Ontological Subject from the list of OSs extracted from the composition, and PMi,j=1 if OSi is in the PCj and 0 otherwise.
The participation matrix is in fact a transformation of information representation from the usual forms of compositions of the ontological subjects, e.g. textual, to numerical matrixes which are easier for processing by computers and specific or predesigned systems of hardware and software.
Furthermore those skilled in the art can store the information of the PMs and other objects of the present invention in different or equivalent data structure forms without using the notion of a matrix. For example each raw of the PM can be stored in data structures such one or more dictionaries, or the PM be stored in one or more list or lists in lists, hash tables, or any other convenient objects of any computer programming languages such as Python, C, Perl, etc. Such practical implementation strategies can be devised by various people in different ways. The detailed description, herein, therefore uses a straightforward mathematical notions and formulas to describe one exemplary way of implementing the methods and should not be interpreted as the only way of formulating the concepts, algorithms, and the introduced mathematical or computer implementable objects, measures, parameters, and variables. Therefore the preferred mathematical formulation here should not be regarded as a limitation or constitute restrictions for the scope and spirit of the invention.
Having built the PM, we then can calculate and construct the co-occurrence matrix by:
C(OSi,OSj)=Ci,j=ƒ(Ri,Rj) (2)
where C is the co-occurrence matrix, Ri and Rj are the ith and jth row of the PM, and ƒ is a predefined function or operator of the two binary vectors Ri, Rj. The function ƒ usually is the inner product or scalar multiplication of the two vectors. The matrix C has the form of:
In this case the matrix C is symmetric and in fact, could be viewed as an adjacency matrix of a weighted undirected graph. It contains useful information that can be used to calculate or estimate the importance of the OS in such graph derived from a composition corresponding to its universe. Importance factor could be simply counts, node centrality measure, etc. More importantly the row of the matrix C shows the association set for each OS with the related association value. Furthermore we can also define a more useful column-normalized association matrix, called A here, that can also be built from C, with entries defined as:
However for some applications, proposed here, such as knowledge navigation and exploration a directed graph which is more like a guiding map is more appropriate and desirable. We consequently introduce the Ontological Subject Map (OSM) which is a multilayer index of OSs configured to position each OS uniquely on a map with connection to its most important associates and multistep routes to all other OSs.
OSM Graphs:
The OSM is essentially a directed (preferably weighted) graph in which each OS is represented by a node as shown in
As seen in the
A dormant node is in fact a mirror of a growing node somewhere in the graph. The corresponding index of the graph contains the information related to the address of originally growing position of the dormant node. In other words, dormant nodes points to their originally growing positions in the graph. However, in
It should be noticed that in
OSM Building Algorithm
In the preferred method, the OSM in
Select a first set of ontological subjects, having at least one member, which have the highest importance factor, e.g. highest occurrence counts. In the map put the first set of OS in the first layer and showing each OS by a node. For each of this first layer OSs form an association set, having a desired number of OSs that have association value of higher than a predetermined threshold, with each first layer OS. This can be done by looking at the adjacency list of each OS in the co-occurrence matrix C or the associated matrix A, and select the first associated sets of ontological subjects, each set associated with each of first layer OSs. Create a second layer of nodes, underneath first layer OSs, and place the associated set of each first layer OS in the second layer underneath its corresponding the first layer OS (also called a parent node here). Each OS, i.e. node, in the associated set placed in the second layer points to its first layer parent node if that OS appears in only one associated set and is not a member of first layer set. In this case the node is called growing or non-dormant. If an OS in the second layer is also a member of first layer set then in the map the parent node points to that OS in the second layer and that OS ultimately address or points to its first appeared position in the first layer. In this case that node in the second layer is called dormant, and would not grow further than the second layer.
If an OS is not a member of first layer but is a member of more than one associated set, then that OS only growing under the parent with which it has the highest association value, points to that parent, and becomes dormant for other associated parents. When the OS becomes dormant, the parents point to that dormant OS and that dormant OS address or points to its growing position in the second layer.
For each of growing OSs of the second layer (called again the parent node as well), form an association set, having a desired number of OSs that have association value of higher than a predetermined threshold, with each of growing OS in the second layer. Create a third layer of nodes, underneath second layer, and place the associated set of each of growing nodes of the second layer OS in the third layer underneath its corresponding the second layer growing OS. Each OS, i.e. node, in the associated set placed in the third layer points to its the second layer parent node if that OS appears in only one associated set and is not a member of the first or the second layer set. If an OS in the third layer is a member of above layers, i.e. the first or the second layer, then in the map the parent node points to that OS in the third layer and that OS ultimately address or points to its first appeared position (growing) in the above layers. In this case that node in the third layer is called dormant, and would not grow further than third layer.
If an OS is not a member of above layers but is a member of more than one associated sets, then that OS only growing under the parent to which it has the highest association value, points to that parent, and becomes dormant for other associated parents. When an OS becomes dormant, the parents point to that dormant OS and that dormant OS address or points to its growing position in the same layer.
For each growing node in the third layer repeat the process and create more layers of the Map until all ontological subjects of the universe found a growing position in the map or until any other predetermined or desired criteria is met. Consequently or at the same time, index the map with appropriate indexing method. The indexing could be done, for example, by storing the adjacency matrix of the map or storing the adjacency list for each growing node in the map. As seen again, an OS can have one growing position but be dormant associates, i.e. dormant node, for many other growing OSs. Therefore dormant nodes are mirrors of growing nodes and essentially pointing to their growing position address in the index or having the same OS number when represented by a matrix.
After building the OSM and the index, we have a directed weighed graph that can be used for knowledge exploration, navigation, and many other applications. More importantly we can intrinsically measure the importance of each OS in the context of its universe.
Adjacency Matrix of OSM Graph
When we consider the OSM as a graph then mathematically we can represent the corresponding graph as: OSM=(OS, E) wherein OS is the set of ontological subject of the universe and E is the set of edges or connections and it is either a growing connection or dormant connections and can be divided as E={ei,j V
The ontological subject map is a directed weighted graph that can be also shown by its adjacency matrix as:
in which, in one preferred embodiment of the invention, we have:
The matrix M is most of the time asymmetric and sparse. The matrix M can further be divided by two adjacency matrix one showing only the growing type connections and another showing the dormant type connections.
OS Power Spectrums
In natural language type reasoning, ontological subjects carry a weight that is inherent in their intrinsic importance as they are the symbols of something in the universe. These symbols have been introduced or invented to name something and to represent something important. The more a subject is discussed by diverse group of people over a long period of time the more its intrinsic importance or power should be.
Power is a good choice for measuring the importance of the ontological subjects since everything goes back to energy and every entity can be represented by some energy value. So in proposing a discipline to the science of knowledge discovery a map that is build based on assigning an energy value or power value to ontological subjects is sensible. Such a map universally shows the connections between the most important things in the universe and sorts them based on their inherent power. The map, therefore, can guide a viewer or user to find subjects of intrinsic value to work on and help them to select an efficient route or trajectory for research and investigation of a subject matter that may lead to valuable results.
Therefore in one embodiment of the present invention we consider the relation between the nodes as a type of energy and power relationship and therefore a node has a power which is coming from its associated set of nodes in the OSM. If we regard the association value as amplitude then we can calculate the power of each OS versus its associated OS as:
Pi=P(OSi)=Σj=1Ngi,j(mi,j)Pj (6)
wherein g is a predefined function and in this embodiment (power/energy relationship between associated OSs) can be given by:
Equation (6) is an eigenvalue equation and the intrinsic power of each OS, in the OSM graph is determined by the stationary solution of the equation (6), i.e. the eigenvector. However to make sure that the equation (6) is computable and has a unique eigenvector, corresponding to its dominant eigenvalue, we can rewrite the equation (6) as:
wherein G is column-normalized matrix with entries gi,j, which is also mostly an asymmetric and spars matrix, I is a N×N matrix with all entries 1, and 0≦γ≦1 is a parameter indicating that some power of each node coming from the rest of the nodes that are not directly associated with that OS. Intuitively and most of the time the value of γ is selected from the interval [0.01, 0.5]. The eigenvalue equation of (8) can be solved numerically, for instance, by the power method and by selecting an initial power vector state P0. The stationary eigenvector therefore, when power method is used, is given by:
The adjacency list of each OS, i.e. each row of the adjacency matrix M or G, can be viewed as the spectrum or power spectrum of each OS versus its associated set of OSs. The power spectrum can be used for quick comparison of different composition to each other and/or to a reference OSM.
Referring to
OSMs for Comparison and Merit Measurement
One of the motives and application of the method and system of the invention is to use the method and system to compare compositions against each other and/or a larger composition and/or a collection of compositions. In doing so, two approaches may be employed alternatively or both at the same time.
One, or the first, approach, which is in fact a special case of the other approach, is to extract the ontological subject set of a first composition, e.g. called OSu1, and build the co-occurrence matrix in u1 for that set, and uses the same set to build the co-occurrence matrix in the partitioned compositions of universe 2, u2.
The universe 2 could be simply another composition or could be a larger universe with more partitioned compositions, such as a collection of compositions, a corpus, or a collection of related compositions obtained from the internet using search engines, etc. In one important case the universe 2 is the repository of the whole internet which in that case the universe 2 is close to our real universe.
Commercial or in house search engine databases can be used to get the co-occurrences counts of each two OSs from the internet. When using internet and search engine, building a co-occurrence matrix could involve simply the “Boolean AND” search for each two OSs and looking at the hit counts. When the number of partitions or the compositions found in the internet, containing both OSs, is large enough, which is usually the case, the hit number is a good approximation of co-occurrence of each two OS in our universe. However for a more certainty in constructing co-occurrence matrix one may chose to download a plurality of composition form the internet and construct the co-occurrence matrix of OSu1 in that collection of compositions which form the universe 2, u2. Using the teachings of the present invention we can then build two OSMs for the ontological subjects derived from u1. One of the OSM is build from the composition of u1 and another is build from composition of another universe say u2. The resulting OSMs denoted as OSMu11 and OSMu21 respectively as shown in
The other approach is to expand the number of OSs beyond the set of OSu1.
To find more compositions containing one or more members of OSu1 we can use internet and search engine, or we can search in a premade database of composition such as large corpuses or collections of diverse compositions. Also, for instance, to find more associated OS for OSu1 and expand the spectrum, we can use the strongest OSs in universe 1, derived from OSMu11, and then search in the internet to get more related compositions from which more associated ontological subjects can be extracted.
Usually one of the universes (often the larger one) is used as the reference universe. The larger universe refers to a universe which has a higher number of ontological subjects, i.e. more knowable objects or subjects. The dimension of the OSM or the resulting matrix M or G is determined by the number of OSs from the larger universe. Hence the matrixes M and G for OSMu11 and OSMu21, and their corresponding stationary vector Pu11 and Pu21 will have the same dimension.
The co-occurrence matrix of the universe with lesser number of OS, will have zero co-occurrence for those OS that do not exist in that universe. For comparison application, the OS axis covers the larger universe OS members. In one particular, but important case, the OS axis could be universal and containing the largest possible number of OS (all the OSs that have existed or known to the present time).
Referring to
Now consider that we want to analyze and assess a composition of universe 1 (u1) in the context of a reference universe 2 (u2). That is to use the ontological subjects of u1 to construct the co-occurrence matrix in both universes. We can, then, build the OSM for each of the universes and construct the matrix M or G and consequently the power vector P for each universe. We now introduce few exemplary measures of merit for a composition of u1, in the context of a reference universe 2, u2. For example one measure of merit or merit parameter can be defined as:
where mp1 is the merit parameter 1, and ∥ ∥ in the norm of a vector. This merit measure is in fact a measure of correctness and substance of the composition of u1 in the context of reference u2. This measure can be readily used for ranking contents, e.g. ranking the contents of web pages or ranking documents in a collection of documents, etc. As seen by those skilled in the art one of the advantages of the power spectrum notion of compositions is the ability to use the well known method of spectral analysis and signal processing in dealing with text compositions or generally content analysis.
The association value matrix A and/or the adjacency matrix M and/or the power matrix G also convey interesting and important information about the content of composition of u1. For instance, another useful set of data related to measures of merit of a composition in the context of the reference universe u2, are obtained by the differential power matrix which is defined as:
Gd=[Gu1−Gu2]. (11)
wherein Gd is the differential power matrix which contains interesting and valuable information about authoritativeness, novelty and/or substance of a composition compared with a reference universe of u1.
The matrix Gd can be represented visually by using, for example, mesh or counter plot from MATLAB® software or any other desirable tools and methods. When the matrix Gd is represented visually, interesting features of the composition of u1 in the context of u2 can be seen. For example when there is a perfect match then the Gd=0 and no bump or intensity difference in the mesh or plot can be seen. However, when Gd≠0 the mesh or plot can show the location and intensity of differences visually, and guide a user to look into these areas for further analysis and investigation. Therefore Gd can point to novelty, new knowledge, or flaws in the composition.
When the reference universe is large enough, the reference universe can be viewed as the contemporary collective knowledge of people as whole or a large group of people expert in a domain of knowledge. For instance, the sum of all rows or columns of the differential matrix, Gd, is an indication of magnitude of general deviation of a composition from the status quo knowledge or collective understanding of the present time about a subject. Alternatively a sum over a row or a column of the differential matrix, Gd, is a measure of local differences and deviation of power and emphasis of each OS, used in the composition, from the collective wisdom or collective knowledge of people about that OS.
Depends on the application, more sophisticated or detailed analysis can be introduced or used without departing from the scope and spirit of the invention. For example one may define another measure of merit or merit parameter as follow:
where mp2 is the second exemplary merit parameter, pu11 and pu21 are the power vector of the universe 1 and 2 respectively, piu11 and piu21 are the power of OSi derived from OSMu11, and OSMu21 respectively, and mi,ju11 and mi,ju21 are the elements of the matrix M corresponding to OSMu11, and OSMu21 respectively. Here mp2≧0 and may be a more accurate measure of similarity and substance than mp1.
Alternative Spectrums and More Merit Measures
More quantitative measures or alternative formulation is possible to envision with minor differences from the method presented in this invention which was explained by the exemplary embodiments.
Specifically the function g in equation (6) can be defined linearly so the elements of matrix G are linear functions of elements of matrix M. In one special case G can be the same as matrix M. Furthermore instead of M, the adjacency matrix of the OSM, one can also use the co-occurrence matrix C or column-normalized association matrix A in equation (6) to (9), to derive another set of similar formulations, or use a different view or interpretation of the OS spectrums of universes. Other types of OS power spectrums or additional calculable parameters and data can also be used, for more comprehensive analysis of compositions and knowledge processing applications. Those skilled in the art can alter the formulations without departing from the scope and spirit of the present invention.
Examplary Applications
The method is based on intrinsic value of subjects in a universal context and therefore a better platform for comparison, ranking, and retrieval applications for the compositions. Therefore, in below few exemplary and non comprehensive applications of the present invention are given.
In many applications it is faster and advantageous to have premade OSMs to be used as references for different compositions.
It should be noticed that in this embodiment the universe contains all the compositions that exist in the internet and therefore the resulting premade OSMs are indicatives of general understanding or distilled state of knowledge about the ontological subjects. When the input list of OS is already classified and contains a group of related OSs, the resulting OSM can also be categorized under the same classifications. Therefore we can have specific premade OSMs for different classes or related OSs. However, when the input OS list is general and large enough, the resulting OSM is also general. The larger the list of input OS, the broader the extent of OSM would be and the closer the OSM would become to underlying realties of the universe. One should expect to have good true knowledge of our universe when the input list of OS contains all existing and conceivable ontological subjects of the world in its largest extent. In this case the resulting OSM is universal and very close to true realties of our universe. A universal OSM can be very instrumental in new knowledge discovery, since the connection of everything in the universe to everything else in the universe is established and revealed.
Application Systems
Many system configurations can be proposed to implement the method and teachings of the present invention that provide a service to users for one or more of the mentioned exemplary applications or many other that were not listed.
For example, according to one embodiment, the computing system that is used to implement the method and teachings of the present invention has more than 1 (one) Giga Byte of RAM capacity and the processing device or units (e.g. the silicon based microprocessor) that can operate with clock speeds of higher than 1 (one) Giga Hertz (e.g. an Intel Pentium 3, Dual core, i3, i5, i7 and Xeon series processors or equivalents or similar from other vendors) are suitable to perform and execute the method once they have been programmed by computer readable instruction/codes/languages or signals and instructed by the executable instructions. Additionally, for instance according to another embodiment of the invention, the computing or executing system includes or has processing device/s such as graphical processing units for visual computations that are for instance, capable of rendering and demonstrating the graphs/maps of the present invention on a display (e.g. LED displays and TV, projectors, LCD, touch screen mobile and tablets displays, laser projectors, gesture detecting monitors/displays, and the like from various vendors, such as Apple, Samsung, Sony, or the like etc.) with good quality (e.g. using a NVidia graphical processing units). Also the methods, teaching and the application programs of the presents invention can be implement by shared resources such as virtualized machines and servers (e.g. VMware virtual machines, Amazon Elastic Beanstalk, e.g. Amazon EC2 and storages, e.g. Amazon S3, and the like etc. Alternatively specialized processing and storage units (e.g. Application Specific Integrated Circuits ASICs, field programmable gate arrays (FPGAs) and the like) can be made and used in eth computing system to enhance the performance and the speed and security of the computing system of performing the methods and application of the present invention.
Alternatively
Moreover several of such computing systems can be run under a cluster, network, cloud, mesh or grid configuration connected to each other by communication ports and data transfers apparatuses such as switches, data servers, load balancers, gateways, modems, internet ports, databases servers, graphical processing units, storage area networks (SANs) and the like etc.
Client server system architecture over networks and internet is well known so that we do not show the exemplary computer architecture and network topology of such client server systems. Accordingly, in
For instance, in this embodiment, a request could be a natural language question which needs one or several statements of the facts as the answer. Alternatively, a user might want to analyze a newly composed composition in the context of much larger reference universe, or a user simply would like to get guidance for researching about a subject matter. In this exemplary embodiment there are functional blocks that identify the type of service that user has requested, and then the request is passed to the principle OS identifier of input information attached to the request. The user provides some content with the request and therefore the rule of the principal OS identifier is to extract the main OS of the content accompanying the request.
Once the type of request was identified the request and its main OS/s get routed to corresponding application engine. The application engine then interacts and communicates with the OS Processing Engine (OSPE) to perform its task and provide the requested information to the user as the output. The outputs corresponding to each service, as expressed in the
The OS Processing Engine (OSPE) is responsible to provide the necessary information and processing functions that is requested by the application engine. The OS processing engine for instance provides the list of associated OS of input OS, either from premade OSMs or by obtaining the related compositions and finding the associated OSs with highest associated value. The OS Processing Engine (OSPE) is capable of building OSM for an input composition on demand. The OSPE will also be able to build an OSM from the repositories that contain a predetermined number of associated OSs to the main OS of the input. Moreover, it can also look for explicit relations between OSs from in house repositories or internet resources. Furthermore, it is capable of verifying the trueness of the statements by statistical analysis of the hunted statements containing one or more of the OSs.
In summary the invention provides method and systems for enhancing new knowledge creation and accelerating the knowledge discovery. The invention can serve knowledge seekers, knowledge creators, inventors, discoverer, as well as general public, by assisting and guiding them to assess their creation, identify their unknowns, and helping them to plan their research trajectory while providing high quality contents related to their working subjects. The method and system, thereby, is instrumental in increasing the speed and efficiency of knowledge creation, retrieval, learning, and problem solving to name a few.
It is understood that the preferred or exemplary embodiments and examples described herein are given to illustrate the principles of the invention and should not be construed as limiting its scope. Various modifications to the specific embodiments could be introduced by those skilled in the art without departing from the scope and spirit of the invention as set forth in the following claims.
Patent | Priority | Assignee | Title |
Patent | Priority | Assignee | Title |
7596549, | Apr 03 2006 | Qurio Holdings, Inc | Methods, systems, and products for analyzing annotations for related content |
20070203693, | |||
20080016072, | |||
20100030552, |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Mar 07 2013 | Hamid Hatami-Hanza | (assignment on the face of the patent) | / |
Date | Maintenance Fee Events |
Sep 02 2018 | M2551: Payment of Maintenance Fee, 4th Yr, Small Entity. |
Sep 17 2022 | M2552: Payment of Maintenance Fee, 8th Yr, Small Entity. |
Date | Maintenance Schedule |
Mar 17 2018 | 4 years fee payment window open |
Sep 17 2018 | 6 months grace period start (w surcharge) |
Mar 17 2019 | patent expiry (for year 4) |
Mar 17 2021 | 2 years to revive unintentionally abandoned end. (for year 4) |
Mar 17 2022 | 8 years fee payment window open |
Sep 17 2022 | 6 months grace period start (w surcharge) |
Mar 17 2023 | patent expiry (for year 8) |
Mar 17 2025 | 2 years to revive unintentionally abandoned end. (for year 8) |
Mar 17 2026 | 12 years fee payment window open |
Sep 17 2026 | 6 months grace period start (w surcharge) |
Mar 17 2027 | patent expiry (for year 12) |
Mar 17 2029 | 2 years to revive unintentionally abandoned end. (for year 12) |